Robot Control Model Decomposition for Interpretable Autonomous Operations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robot control devices lack transparency in showing users what operations a trained model has been trained for, making it difficult for users to confirm the autonomous operation of robots.
Innovation Solution
A robot control device that includes a trained model, a control data acquisition section, an operation label storage section, and a base trained model combination information acquisition section, which acquires and outputs operation labels to explain the combination of operations trained by the model, increasing transparency and user satisfaction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If a trained model is built by machine learning to enable autonomous robot operation, then the robot's automation capability is improved, but the transparency of the model's operations deteriorates
Solution Approach 1:
The patent introduces base trained models as intermediary components that decompose the complex trained model into interpretable operation units. Each base trained model corresponds to a specific operation type and serves as a mediator between the black-box trained model and the user, providing transparency through operation labels and combination information without compromising the autonomous operation capability.
Solution Approach 2:
The patent segments the trained model into multiple base trained models, each responsible for a specific operation. This segmentation allows the complex autonomous operation to be broken down into understandable components, where each base model's function and combination can be explained through operation labels, thus improving transparency while maintaining automation.
2Loss of information
If the trained model structure is simplified for easier interpretation, then the transparency is improved, but the complexity of the model combination increases
Solution Approach 1:
The patent creates base trained models with universal functionality where each base model can handle multiple aspects of operation interpretation. The operation labels serve as a universal interface that can describe different operations consistently, and the combination information structure provides a unified method to represent complex model compositions, thereby managing complexity while maintaining transparency.
Data Source
AI summary
A robot control device includes: a trained model built by being trained on work data; a control data acquisition section which acquires control data of the robot based on data from the trained model; base trained models built for each of a plurality of simple operations by being trained on work data; an operation label storage section which stores operation labels corresponding to the base trained models; a base trained model combination information acquisition section which acquires combination information when the trained model is represented by a combination of a plurality of the base trained models, by acquiring a similarity between the trained model and the respective base trained models; and an information output section which outputs the operation label corresponding to each of the base trained models which represent the trained model.